[Libreoffice-bugs] [Bug 126202] Writer: Insert Hyperlink dialogue: can't enter or edit the URL (GTK3)

[bugzilla-daemon at bugs.documentfoundation.org]

https://bugs.documentfoundation.org/show_bug.cgi?id=126202

--- Comment #12 from Adalbert Hanßen <adalbert.hanssen at gmx.de> ---
(In reply to David Gerard from comment #10)
> I compile it myself. On this box, my --prefix=/home/fun/bin/libreoffice so I
> start it from the command line with this command to get gtk2:
> 
> SAL_USE_VCLPLUGIN=gtk
> /home/fun/bin/libreoffice/lib/libreoffice/program/soffice

David, this is a very valuable finding because it will let someone who really
understands those basic functions from gtk, qt and the like direct to really
trap the bug. The outcome of this might be "not my bug, it is in gtk". But this
would not really eliminate it, because such bugs can trigger more severe
results than just not being able to set the URL for a hyperlink!

There would not be different versions of gtk unless there are differences
between gtk2 and gtk3. Therefore there may be more than one applications
affected by the bug.

Let me guess, the possible root cause is probably is one of the following:

1. a calling interface has been changed, e.g. by introducing additional
parameters and dealing with variable length calling interfaces is not properly
implemented everywhere (going through the steps to edit a hyperlink in this
case will lead to the one affected - unfortunately among many others!)

or 

2. by allowing different ranges for values of parameters. If defensive
programming style limited the range of such a parameter before, it has worked
before but does not work any longer, after the extended value range is used for
different situations and the "limit to what is allowed" strategy has been
removed to allow such extended use of existing parameters, but with different
values.

or

3. calling some function required some preconditions being met which are no
longer guaranteed, e.g.

3a. in the old code no distinction was necessary between two different use
cases of some code segment which existed in multiple similar instances which
have been unified. The user of the old code did not have to care about a
preconditions because it has been introduced in the unification those similar
instances. Therefore one should look if any preconditions were introduced later
(hopefully they are documented!).


3b. because some defensive programming code has been disabled which formerly
coerced the precondition to be met by substituting what was missing before. But
later someone wanted to speed up things and removed or disabled such
"unnecessary stuff" because it was clearly said that the precondition must have
been met before calling the function in question. Hopefully coercing the
precondition has not been completely removed from the code but can be
re-activated by setting proper compilation flags and hopefully they can be
reactivated in such a way to eject debugging information that a superfluous
safety measure really had to be taken because it was necessary in this case... 

or

4. some of the other rules in
https://developer.gnome.org/gtk3/stable/gtk-migrating-2-to-3.html has not been
observed

Anyway, this bug which may (or may not) "live" in the gtk region of the
software, is an interesting one as it might go unnoticed in many use cases and
in the worst case it might be used to make malicious use of functions. It
should be corrected and reasonable thought should be spent in checking if there
are other similar cases which might trigger the same bug.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/libreoffice-bugs/attachments/20190706/0c8d279d/attachment.html>